Workload forecasting and capacity planning are fundamental in provisioning of web services and applications. It is common for workload parameters to be analyzed “after the fact”, for example, to compensate for workload levels already being experienced by a particular web server. As a result, such a workload analysis of a system does not provide any insight into the level or location of activity that a network system is likely to experience in the future. However, forecasting user activity patterns is crucial for efficiently deploying and organizing network server resources. Knowledge regarding the relationships between network users can also be used in devising ways to handle the activity patterns that a network system is likely to experience in the future.